CN113505533A

CN113505533A - Equipment health state prediction method and device

Info

Publication number: CN113505533A
Application number: CN202110761529.5A
Authority: CN
Inventors: 张燧; 徐少龙
Original assignee: Ennew Digital Technology Co Ltd
Current assignee: Ennew Digital Technology Co Ltd
Priority date: 2021-07-06
Filing date: 2021-07-06
Publication date: 2021-10-15

Abstract

The invention relates to the technical field of equipment health management, and provides a method and a device for predicting the health state of equipment. The method comprises the following steps: collecting corresponding real-time data in a preset time according to the use characteristics of equipment, acquiring marking data based on the collected real-time data, and establishing a hidden Markov model; acquiring a state transition expected number based on the labeling data; obtaining a new state transition expectation number based on the state transition expectation number; and training a prediction model based on the new state transition expectation number to obtain a target prediction model, and calculating the residual service time of the equipment based on the target prediction model. The method and the device can reduce maintenance cost, ensure normal operation and avoid the crisis that the whole comprehensive energy system is paralyzed due to the failure of one device by adopting the hidden Markov model and the Monte Carlo to calculate the residual service time of the device.

Description

Equipment health state prediction method and device

Technical Field

The invention relates to the technical field of equipment health management, in particular to a method and a device for predicting the health state of equipment.

Background

In the comprehensive energy system, the health degree of a large amount of equipment is often damaged due to long-term work, environmental change, frequent start and stop and other reasons, and when the regular maintenance time is not reached, the equipment breaks down, so that the maintenance cost is increased, the normal operation is influenced, and even the whole comprehensive energy system is paralyzed to cause serious loss.

Therefore, an assessment of the health of the device is extremely necessary.

Disclosure of Invention

In view of this, the disclosed embodiments of the present invention provide a method and an apparatus for predicting a health status of a device, so as to solve a problem that a remaining usage time of the device in an integrated energy system in the prior art cannot be predicted.

In a first aspect of the disclosed embodiments of the present invention, a method for predicting a health status of a device is provided, including:

collecting corresponding real-time data in preset time according to the use characteristics of equipment, wherein the real-time data at least comprises one of equipment operation age, historical maintenance data, current operation data and energy consumption data;

acquiring marking data based on the collected real-time data, and establishing a hidden Markov model;

acquiring a state transition expected number based on the labeling data;

obtaining a new state transition expectation number based on the state transition expectation number;

and training a prediction model based on the new state transition expected number to obtain a target prediction model, and calculating the residual service time of the equipment based on the target prediction model.

In a second aspect of the disclosed embodiments of the present invention, an apparatus for predicting health status of a device is provided, including:

the collecting module is configured to collect corresponding real-time data in a preset time according to the use characteristics of the equipment, and the real-time data at least comprises one of equipment operation years, historical maintenance data, current operation data and energy consumption data;

a labeling module configured to obtain labeling data based on the collected real-time data and establish a hidden Markov model;

migrating the desired module; configured to obtain a state transition expectation number based on the annotation data;

a new migration expectation module configured to obtain a new state migration expectation number based on the state migration expectation number;

and the calculation module is configured to perform prediction model training based on the new state transition expected number, obtain a target prediction model and calculate the residual service time of the equipment based on the target prediction model.

In a third aspect of the disclosed embodiments of the present invention, there is provided a computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the steps of the above method when executing the computer program.

In a fourth aspect of the disclosed embodiments, a computer-readable storage medium is provided, in which a computer program is stored, which, when executed by a processor, implements the steps of the above-mentioned method.

Compared with the prior art, the embodiment disclosed by the invention has the beneficial effects that: according to the using characteristics of the equipment, the embodiment of the invention collects corresponding real-time data in preset time; establishing a hidden Markov model, obtaining marking data, a state transition expected number and a new state transition expected number, training the new state transition expected number to obtain a target prediction model, and finally calculating the residual service time of the equipment by adopting the target prediction model. The method and the device for calculating the residual service time of the equipment have the advantages that the hidden Markov model and Monte Carlo simulation calculation are combined, maintenance cost can be reduced, normal operation is guaranteed, and the danger that the whole comprehensive energy system is paralyzed due to the fact that one piece of equipment breaks down is avoided.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the drawings used in the embodiments or the prior art descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present disclosure, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.

FIG. 1 is a flow chart of a method for predicting health status of a device according to an embodiment of the present disclosure;

FIG. 2 is a flowchart illustrating the process of obtaining annotated data in a method for predicting health status of a device according to an embodiment of the disclosure;

fig. 3 is a flowchart of obtaining a state transition expectation number in a method for predicting a health state of a device according to an embodiment of the disclosure;

FIG. 4 is a flowchart illustrating obtaining a new expected number of state transitions in a method for predicting a health status of a device according to an embodiment of the disclosure;

FIG. 5 is a flowchart illustrating a method for predicting remaining usage time of a device according to an embodiment of the disclosure;

FIG. 6 is a block diagram of an apparatus health status prediction device according to an embodiment of the disclosure;

FIG. 7 is a schematic diagram of a computer device according to an embodiment of the disclosure.

Detailed Description

In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the disclosed embodiments of the invention. It will be apparent, however, to one skilled in the art that the present disclosure may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present disclosure with unnecessary detail.

A method and apparatus for predicting health status of a device according to embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.

Fig. 1 is a flowchart of a method for predicting a health status of a device according to an embodiment of the disclosure. As shown in fig. 1, the device health prediction method includes:

and S11, collecting corresponding real-time data in a preset time according to the use characteristics of the equipment, wherein the real-time data at least comprises one of equipment operation age, historical maintenance data, current operation data and energy consumption data.

According to the use characteristics of equipment, corresponding real-time data are collected within a preset time period, due to some uncertain factors, data are often lost or abnormal, and similar conditions can be combined with a regression classification method according to expert experience to carry out numerical interpolation, abnormal value removal and other operations. Among these, regression-type methods, for example, least squares, polynomial regression, and the like. And the real-time data at least comprises one of equipment operation age, historical maintenance data, current operation data and energy consumption data. The current operation Data is collected in a scada (Supervisory Control And Data Acquisition, Data Acquisition And monitoring Control system), while the equipment operation age, historical maintenance Data And energy consumption Data are generally collected in a ledger record, And only a company which strictly manages the Data in a standard manner can directly collect the Data in the scada.

And S12, acquiring marking data based on the collected real-time data, and establishing a hidden Markov model.

Fig. 2 is a flowchart of obtaining annotation data in a method for predicting a health status of a device according to an embodiment of the disclosure. As shown in fig. 2, S12 includes:

and S121, defining state parameters based on the real-time data.

The hidden Markov model carries out state parameter defining processing on the real-time data collected in the preset time to obtain the real-time data after the state of the parameter is defined. The state parameter is defined according to historical maintenance data in the real-time data, and the historical maintenance data records various types of data in detail, so that data can be relied on by the data.

And S122, classifying the real-time data with the state parameters defined according to a preset segmentation mode to obtain labeled data, wherein the preset segmentation mode is to divide the real-time data with the state parameters defined into a plurality of data intervals, and each data interval corresponds to different data types. Wherein defining the state parameters is followed by classifying for current operating parameters in the real-time data.

For example, the real-time data after the state parameter is defined is classified, for example, the real-time data after the state parameter is defined is classified into four classes, and the four classes are labeled as good, medium, and bad, respectively, so that the obtained data is labeled data.

For another example, the range of the real-time data after the parameter state is defined is classified, for example, the real-time data after the parameter state is defined is classified into three classes, and the three classes are labeled as a good class, a medium class and a poor class, respectively, and then the obtained data is the labeled data.

For another example, the ranges of the real-time data after the parameter states are defined are classified, for example, the real-time data after the parameter states are defined are classified into five classes, and the five classes are labeled as good, medium, and bad, respectively, and then the obtained data is labeled data.

And S123, establishing a hidden Markov model based on the marking data.

λ＝(A,B,π)---(1)

Wherein, lambda represents hidden Markov model, A is state transition probability distribution, B is continuous current operation parameter value probability distribution, and pi is initial state distribution,

A＝{a_ij}---(2)

a_ij＝P[q_t+1＝S_j|q_t＝S_i]---(3)

wherein, a_ijIs the probability from i state time to j state time, i and j are positive integers, P is the probability from t to t +1 time, S_jIs a state at a certain time, S_iIs a state at a certain time, q_tIs a state parameter at time t, q_t+1Is the state parameter at time t +1,

B＝{bj(O)}---(4)

wherein, b_jFor one of the continuous current operating parameter valuesProbability, m is one of the different data, K is the number of participating mixing parameters, C_jmFor mixing parameters, N (O | μ)_jm,σ_jm) Is a Gaussian mixture model with Gaussian distribution, O is the current operating parameter, mu is the expectation number, and sigma is the standard deviation.

π＝{π_i}---(6)

Wherein, A and pi can be obtained from field experience and belong to known variables.

S13, obtaining the expected number of state transitions based on the marking data.

Fig. 3 is a flowchart for obtaining a state transition expectation number in a method for predicting a health state of a device according to an embodiment of the disclosure. As shown in fig. 3, S13 includes:

s131, based on the hidden Markov model and the current operation data, obtaining the probability from the current time state to the next time state at a certain time,

ε_t(i,j)＝P[q_t＝S_i,q_t+1＝S_j|O，λ]---(7)

equation (7) is a special case of equation (3), namely the transition probability from state i to state j at time t. The establishment of the hidden Markov model can establish the correlation between O and A, because S_jIs a state at a certain time, S_iAnd when the current operation data O is substituted into the state at the moment i, the transition probability from the state i to the state j at the moment t is obtained.

Wherein epsilon_t(i, j) is the transition probability from state i to state j at time t, t being one time, t +1 being the next time to t.

And S132, obtaining the expected number of state transition based on the probability from the state at the previous moment to the state at the next moment.

Similarly, based on the method of the probability from the previous state to the next state, the probability from several previous states to the next state can be calculated, and thus, the expected number of state transitions can be obtained.

Wherein ε (i, j) is the slave state S_iTo state S_jT is a set of T, and 1T T.

S14, a new expected number of state transitions is obtained based on the expected number of state transitions.

And obtaining the new state transition expectation number by using the original time state expectation number and the new time state expectation number.

Fig. 4 is a flowchart for obtaining a new expected number of state transitions in a method for predicting a health status of a device according to an embodiment of the disclosure. As shown in fig. 4, S14 includes:

and S141, obtaining a Bayesian information criterion minimum threshold value based on the state transition expectation number.

BIC＝L*InT-2InP(O|λ)---(9)

Since L is a constant value, T is a constant value, and λ is determined by A, B, and π in equation (9), and is also a constant value, the minimum threshold of Bayesian information criterion can be obtained when newly collected current operating data is used.

Where L is the number of parameters, T is time, and P (O | λ) is the probability of the current running data under a given model condition, BIC (Bayesian Information Criterion). The embodiment of the invention adopts BIC, can estimate the partially unknown state by subjective probability under incomplete information, then corrects the occurrence probability by a Bayesian formula, and finally makes an optimal decision by using an expected value and the correction probability.

And S142, matching the real-time data with the minimum threshold of the Bayesian information criterion to obtain a new state transition expected number.

ε_new(i,j)＝ε_model(i,j)-ε_data(i,j)-----(10)

Wherein epsilon_dataFor new time state expectation number, epsilon_modelThe state expectation number, epsilon, of the last moment of the new moment_newThe desired number is migrated for the new state.

S15, based on the new state migration number, carrying out prediction model training to obtain a target prediction model, and based on the target prediction model, calculating the residual service time of the equipment,

the new state transition expectation numbers form a large number of data sets, the prediction model is trained on the basis of the large number of data sets, the target prediction model is obtained, and the residual service time of the equipment is calculated on the basis of the target prediction model.

The calculation process relates to Monte Carlo simulation calculation and a Monte Carlo model, the new state transition expectation number and the real-time data after the state parameters are defined are input, and the residual service time of the equipment is output, wherein the residual service time can be presented in a time form and a probability or time form.

Fig. 5 is a flowchart illustrating a method for predicting remaining usage time of a device according to an embodiment of the disclosure. As shown in fig. 5, S15 includes:

and S151, downloading the model to be trained from the server aiming at different equipment, wherein each equipment at least corresponds to one prediction model.

And acquiring a new state transition expected number based on the state transition expected number, and acquiring a new state transition probability distribution matrix based on the new state transition expected number.

Wherein A is_newFor a new state transition probability distribution matrix, ε_new(i, j) is the next time state probability.

Different devices are arranged locally, the server is provided with corresponding prediction models to be trained, the corresponding prediction models to be trained are downloaded from the server, and each device corresponds to more than or equal to 1 prediction model.

And S152, training the prediction model corresponding to each equipment by adopting the new state transition expectation number corresponding to the equipment, and uploading the parameter data which is obtained after training and needs to be updated to a server.

The server trains a large number of data sets comprising the new state transition expectation numbers, each device utilizes the corresponding new state transition expectation numbers to train a prediction model corresponding to the device, parameter data needing to be updated are obtained, and the parameter data needing to be updated are uploaded to the server. The uploading of the parameter data needing to be updated is realized by means of encryption gradient. After training, a new expected number of state transitions can be obtained.

Based on the current time state transition expectation number and the new time state transition expectation number, the RUL (Remaining Useful Life) is obtained through Monte Carlo simulation calculation. The Monte Carlo simulation process repeatedly takes the next state as the current state, and calculates the state of the equipment fault according to the expected number of state transitions, wherein the number of the experienced state transitions is the RUL value. The method disclosed by the embodiment of the invention adopts Monte Carlo simulation calculation, can accurately calculate the residual service life of the equipment, can greatly reduce the loss caused by system crash, and improves the operation reliability of the system.

And S153, downloading a target prediction model from the server, wherein the target prediction model is obtained by updating the prediction model by the server based on the parameter data uploaded by each device.

After each device receives the feedback of the corresponding target prediction model obtained by the server, the device model of the current state of the device is updated by respectively downloading the corresponding updated target prediction model.

Illustratively, the server a divides the data set into three devices, namely a device a1, a device a2 and a device a3, for example, the devices respectively train respective models by using a data set B1, a data set B2 and a data set B3 distributed to the devices themselves, after the respective models are trained, the devices a1, a device a2 and a device a3 respectively upload parameters c1, c2 and c3 which need to be updated to the server a through encryption gradients, the server a aggregates updated model parameters c1, c2 and c3 of the devices and respectively updates the parameters c1, c2 and c3 to the corresponding prediction models, respectively obtains target prediction models B1, B2 and B3, respectively feeds back the updated target prediction models B1, B2 and B3 to the corresponding devices, and downloads the target prediction models from the server a1, the device a2 and the device a3 respectively. The embodiment of the invention adopts a fusion mode that real-time data and fusion parameter standards are matched one by one, and finally achieves the effect that the model is effective under the overall situation, namely a combined framework.

And S154, aiming at different equipment, calculating by adopting a target prediction model corresponding to the equipment to obtain the residual service time of the equipment.

Specifically, the minimum threshold of the bayesian information criterion for predicting the remaining service life of the device is determined according to a preset threshold, and if the threshold is greater than the threshold in the trained data set, the unhealthy state is determined to occur. And on the basis of the trained data set, adopting Monte Carlo calculation to predict the residual service time when the unhealthy state appears in the data set. According to the technical scheme provided by the embodiment of the invention, the collected real-time data is placed in the distribution data set under the joint frame to be trained to different devices in a mode of combining the hidden Markov model and Monte Carlo calculation, and finally the residual service time of the degradation state of the devices is calculated. The embodiment of the invention can help prevent maintenance errors and avoid equipment operation under unsafe conditions.

According to the using characteristics of the equipment, the embodiment of the invention collects corresponding real-time data in preset time; establishing a hidden Markov model, obtaining marking data, a state transition expected number and a new state transition expected number, training the new state transition expected number to obtain a target prediction model, and finally calculating the residual service time of the equipment by adopting the target prediction model. The method and the device for calculating the residual service time of the equipment have the advantages that the hidden Markov model and Monte Carlo simulation calculation are combined, maintenance cost can be reduced, normal operation is guaranteed, and the danger that the whole comprehensive energy system is paralyzed due to the fact that one piece of equipment breaks down is avoided.

All the above optional technical solutions may be combined arbitrarily to form optional embodiments of the present application, and are not described herein again.

The following are embodiments of the disclosed apparatus that may be used to perform embodiments of the disclosed methods. For details which are not disclosed in the embodiments of the apparatus disclosed in the present invention, refer to the embodiments of the method disclosed in the present invention.

Fig. 6 is a block diagram of an apparatus health status prediction device according to an embodiment of the disclosure. As shown in fig. 6, the device health prediction apparatus includes:

the collecting module 61 is configured to collect corresponding real-time data in a preset time according to the use characteristics of the equipment, wherein the real-time data at least comprises one of equipment operation years, historical maintenance data, current operation data and energy consumption data;

a labeling module 62 configured to obtain labeling data based on the collected real-time data and to build a hidden markov model;

a migration expectation module 63; configured to obtain a state transition expectation number based on the annotation data;

a new migration expectation module 64 configured to obtain a new state migration expectation number based on the state migration expectation number;

a calculation module 65 configured to perform prediction model training based on the new state transition expectation number, obtain a target prediction model, and calculate the remaining usage time of the device based on the target prediction model.

It should be understood that, the sequence numbers of the steps in the foregoing embodiments do not imply an execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present disclosure.

Fig. 7 is a schematic diagram of a computer device 7 provided by the disclosed embodiment of the invention. As shown in fig. 7, the computer device 7 of this embodiment includes: a processor 701, a memory 702, and a computer program 703 stored in the memory 702 and executable on the processor 701. The steps in the various method embodiments described above are implemented when the computer program 703 is executed by the processor 701. Alternatively, the processor 701 implements the functions of each module/unit in each device embodiment described above when executing the computer program 703.

Illustratively, the computer program 703 may be partitioned into one or more modules/units, which are stored in the memory 702 and executed by the processor 701 to accomplish the present disclosure. One or more modules/units may be a series of computer program instruction segments capable of performing specific functions, which are used to describe the execution of the computer program 703 in the computer device 7.

The computer device 7 may be a desktop computer, a notebook, a palm computer, a cloud server, or other computer devices. The computer device 7 may include, but is not limited to, a processor 701 and a memory 702. Those skilled in the art will appreciate that fig. 7 is merely an example of a computer device 7 and does not constitute a limitation of the computer device 7 and may include more or fewer components than shown, or some components may be combined, or different components, e.g., the computer device may also include input output devices, network access devices, buses, etc.

The Processor 701 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic device, discrete hardware component, or the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.

The storage 702 may be an internal storage unit of the computer device 7, for example, a hard disk or a memory of the computer device 7. The memory 702 may also be an external storage device of the computer device 7, such as a plug-in hard disk provided on the computer device 7, a Smart Media Card (SMC), a Secure Digital (SD) Card, a Flash memory Card (Flash Card), and the like. Further, the memory 702 may also include both an internal storage unit of the computer device 7 and an external storage device. The memory 702 is used to store computer programs and other programs and data required by the computer device. The memory 702 may also be used to temporarily store data that has been output or is to be output.

It will be apparent to those skilled in the art that, for convenience and brevity of description, only the above-mentioned division of the functional units and modules is illustrated, and in practical applications, the above-mentioned function distribution may be performed by different functional units and modules according to needs, that is, the internal structure of the apparatus is divided into different functional units or modules, so as to perform all or part of the functions described above. Each functional unit and module in the embodiments may be integrated in one processing unit, or each unit may exist alone physically, or two or more units are integrated in one unit, and the integrated unit may be implemented in a form of hardware, or in a form of software functional unit. In addition, specific names of the functional units and modules are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present application. The specific working processes of the units and modules in the system may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In the above embodiments, the descriptions of the respective embodiments have respective emphasis, and reference may be made to the related descriptions of other embodiments for parts that are not described or illustrated in a certain embodiment.

Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present disclosure.

In the embodiments provided in the present disclosure, it should be understood that the disclosed apparatus/computer device and method may be implemented in other ways. For example, the above-described apparatus/computer device embodiments are merely illustrative, and for example, a division of modules or units, a division of logical functions only, an additional division may be made in actual implementation, multiple units or components may be combined or integrated with another system, or some features may be omitted, or not implemented. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.

Units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present disclosure may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

The integrated modules/units, if implemented in the form of software functional units and sold or used as separate products, may be stored in a computer readable storage medium. Based on such understanding, all or part of the flow in the method of implementing the embodiments of the present disclosure may also be implemented by a computer program instructing related hardware, where the computer program may be stored in a computer readable storage medium, and when the computer program is executed by a processor, the steps of the above-described method embodiments may be implemented. The computer program may comprise computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, usb disk, removable hard disk, magnetic disk, optical disk, computer Memory, Read-Only Memory (ROM), Random Access Memory (RAM), electrical carrier wave signals, telecommunications signals, software distribution medium, and the like. It should be noted that the computer readable medium may contain suitable additions or additions that may be required in accordance with legislative and patent practices within the jurisdiction, for example, in some jurisdictions, computer readable media may not include electrical carrier signals or telecommunications signals in accordance with legislative and patent practices.

The above examples are only for illustrating the technical solutions disclosed by the present invention, and are not limiting; although the present disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art will appreciate that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not substantially depart from the spirit and scope of the embodiments disclosed herein and are intended to be included within the scope of the present disclosure.

Claims

1. A method for predicting a health state of a device, comprising:

acquiring a state transition expected number based on the labeling data;

2. The method according to claim 1, wherein collecting the corresponding real-time data in a predetermined time according to the device usage characteristics comprises:

collecting current operating data of the device by a data acquisition and monitoring control system;

and recording the equipment operation age, historical maintenance data and energy consumption data of the equipment through the machine account.

3. The method of claim 1, wherein obtaining annotation data and building a hidden markov model based on the collected real-time data comprises:

defining a state parameter based on the real-time data;

classifying the real-time data after the state parameters are defined according to a preset segmentation mode to obtain labeled data, wherein the preset segmentation mode is to divide the real-time data after the state parameters are defined into a plurality of data intervals, and each data interval corresponds to different data types;

establishing a hidden Markov model based on the marking data, wherein the hidden Markov model is as follows:

λ＝(A,B,π)

wherein, lambda represents hidden Markov model, A is state transition probability distribution matrix, i and j are positive integers, B is continuous current operation parameter value probability distribution, and pi is initial state distribution.

4. The method of claim 1, wherein obtaining the expected number of state transitions based on the annotation data comprises:

obtaining the probability from the current time state to the next time state at a certain moment based on the hidden Markov model and the current operation data,

ε_t(i,j)＝P[q_t＝S_i,q_t+1＝S_j|O，λ]

wherein S is_iAt time t, S_jIs the state at time t +1, epsilon_t(i, j) is time t from S_iTo S_jI is the ordinal number of the state, is a positive integer, t is a moment, and t +1 is the next moment of t;

obtaining the expected number of state transitions based on the probability from the state of the previous moment to the state of the next moment,

where ε (i, j) is the expected number of state transitions.

5. The method of claim 1, wherein obtaining a new expected number of state transitions based on the expected number of state transitions comprises:

obtaining a Bayesian information criterion minimum threshold value based on the state transition expectation number;

and matching the real-time data with the minimum threshold of the Bayesian information criterion to obtain a new state transition expected number.

6. The method of claim 5, wherein the Bayesian information criterion minimum threshold expression is expressed as:

BIC＝L*InT-2InP(O|λ)

wherein, L is the number of parameters, T is the set of each moment, P (O | lambda) is the probability of the current operation data under the given model condition, and BIC is the minimum threshold of the Bayesian information criterion.

7. The method of claim 5, wherein the new state transition expectation number is calculated as follows:

ε_new(i,j)＝ε_model(i,j)-ε_data(i,j)

wherein epsilon_dataFor new time state expectation number, epsilon_modelThe expected number for the original state.

8. The method of claim 7, wherein the new state transition probability distribution matrix comprises:

9. The method of claim 1, wherein the performing predictive model training based on the new state transition expectation, obtaining a target predictive model, and calculating the remaining usage time of the device based on the target predictive model comprises:

downloading models to be trained from a server aiming at different devices, wherein each device at least corresponds to one prediction model;

training a prediction model corresponding to each equipment by adopting the new state transition expectation number corresponding to the equipment, and uploading parameter data which is obtained after training and needs to be updated to a server;

downloading a target prediction model from a server, wherein the target prediction model is obtained after the server updates the prediction model based on parameter data uploaded by each device;

and aiming at different devices, calculating by adopting a target prediction model corresponding to the devices and a Monte Carlo simulation method to obtain the remaining service time of the devices.

10. An apparatus for predicting a health state of a device, comprising: